Identifying AI Opportunities Within the Strategy

With the North Star in place, the work changes shape. The question is no longer “what is this business and where is it going” — that has been answered. The question now becomes: where, specifically, might AI advance any of it?

This is where the framework starts producing AI-shaped artifacts for the first time. The temptation at this point is enormous: to walk in with a list of AI use cases, map them onto the strategy, and call it opportunity identification. That inverts the order. The strategy generates the opportunity list; the opportunity list does not get back-fitted onto the strategy.

This chapter explains how to surface AI opportunities the right way — anchored to the strategy from Chapter 4, broad enough to support real comparison, and held lightly enough that nothing gets prematurely committed to.

Three categories worth walking

AI opportunities tend to live in three places, and a thorough opportunity surface walks all three. Each category names not just a domain but a direction of value.

The first is operational efficiency. This is the inside of the business — workflows, decisions, processes, the things that happen between the moment work starts and the moment it ends. The category is about doing existing work faster, cheaper, or with fewer errors. Slow handoffs, repetitive document work, decisions that wait on aggregation, quality checks that depend on attention. Operational efficiency is where AI opportunities are easiest to spot, because the friction is most visible to the people inside the company. Most AI adoption efforts surface five or six of these in the first hour.

The second is customer delight. This is the outside of the business — every moment a customer interacts with the company, from awareness through onboarding through service through renewal. The category is named for the bar these opportunities have to clear: would a customer actually be glad we did this? AI opportunities in customer-facing contexts are easy to imagine and dangerous to misjudge. A chatbot that frustrates is worse than no chatbot. An automation that erodes trust costs more than it saves. The category is named delight deliberately, to filter for the AI deployments customers would actually celebrate rather than tolerate.

The third is novel capabilities. This is the most abstract of the three. Novel capabilities are not about a specific workflow or a specific customer moment; they are about something the organization currently cannot do, or cannot do at scale, that AI might make newly possible. The ability to personalize at scale. The ability to detect patterns in unstructured data the company has accumulated but never used. The ability to make certain kinds of decisions faster than the current organizational tempo allows. Novel capabilities are often the highest-leverage and the hardest to spot, because they ask the team to see beyond what the business already does.

A surface that covers only the first category produces a thin opportunity set — full of process automation, light on the kinds of opportunities that would actually move the strategy. A surface that covers all three produces a richer set and a better prioritization conversation in the chapter that follows.

How the surface gets generated

The most useful method is to walk the strategy directly. Each of the four anchor questions from Chapter 4 generates a different angle on the opportunity space.

What is the business? This question generates an operational walkthrough. Map the business model — the activities, resources, and value flows that produce the company’s output. At each step, ask: is there a moment here where AI might reduce friction, accelerate the work, or enable something the current process cannot do?

What makes it work? This question generates a capability walkthrough. The thing that makes the business distinctive — its competitive engine — is the thing AI could most powerfully accelerate, or most dangerously erode. Are there opportunities to reinforce the engine? Are there places AI deployment would, ironically, undermine the very thing that makes the business work? Both kinds of opportunity surface here, and both are worth naming.

Where is it going? This question generates a future-state walkthrough. Look at the picture of what the organization is trying to become. What would the future-state version of the business need to do that the current version cannot? AI opportunities here are often capability-shaped — things that would have to be true for the future business to work at all.

What threatens it? This question generates a defensive walkthrough. Where are the vulnerabilities? Are there threats AI could mitigate — emerging competitors, capability gaps, structural shifts? Are there threats AI itself introduces — disintermediation, commoditization of the company’s distinctive work? Both directions matter.

Walking each angle deliberately produces breadth. Walking only one or two — usually the operational walkthrough, because it is the easiest — produces a narrow set that biases the prioritization conversation toward incremental cost savings rather than strategic acceleration.

Appendix A lists ten use case archetypes that AI applications most often draw from — data structuring, document synthesis, classification and tagging, pattern detection, content generation, personalization at scale, conversational interfaces, workflow automation, decision support, monitoring and alerting. The archetypes are a useful reference during this work, especially when a team feels stuck. They help spot opportunities the team would not naturally name on its own.

What to write down

Each opportunity in the surface gets two or three sentences. What it is. Which strategic priority it serves. A first-pass thought on impact (small, medium, large) and feasibility (quick, moderate, hard). Nothing more. The opportunity is a candidate, not a proposal. Detailed scoping comes later, and only for the candidates that survive the prioritization in Chapter 6.

A working surface for an organization at this stage typically contains fifteen to thirty candidates spread across the three categories. Fewer than that suggests the surface is too narrow — usually because only one walkthrough was done, or because the team is implicitly self-censoring opportunities they think are too ambitious. More than that suggests a different problem: opportunities that are not actually anchored to the strategy, or duplicate opportunities phrased differently, or ideas that should be filtered before they enter the surface.

The discipline is to keep each candidate light enough that the team is willing to throw it away. A candidate written in three sentences can be killed in the next conversation. A candidate written as a one-page proposal develops political weight that makes it harder to drop, even if the comparison reveals it should not survive.

The premature-commitment trap

The most common failure at this stage is finding a single compelling opportunity early in the surface and short-circuiting the rest of the work. Someone surfaces an idea that excites the room. The conversation pivots. Time gets spent developing that one idea rather than completing the surface.

This is the moment where the framework’s discipline gets tested. A genuinely good opportunity does not need to be committed to right now. It will still be a good opportunity once the rest of the surface has been generated and a real comparison can be made. An opportunity that loses interest under comparison was never as good as it looked. Skipping the comparison costs more than it saves; the organization commits to a project that may turn out to have been the second-best use of its capacity, with no way to know what was foregone.

The right move when this happens is to capture the exciting opportunity on the surface — three sentences, like the others — and keep walking. It will get its turn in Chapter 6.

Holding the surface lightly

The output of this work is a list — fifteen to thirty AI candidates, each anchored to a strategic priority, each described briefly, none of them yet committed to. It is not a deck. It is not a proposal package. It is a working artifact that goes into the prioritization conversation in the next chapter.

Done well, the surface is light enough that the team can still revise it. Done poorly, it becomes a polished document nobody can edit anymore. The right mindset is the same one that runs through the whole framework: this is what we currently believe, parts of it will be wrong, and the next stage will sharpen it.

The next chapter takes the surface and turns it into a small set of starting projects.